Abstract
Some researchers have suggested Golden Eagle (Aquila chrysaetos) populations may be declining in at least part of their range (Bittner and Oakley 1999, Leslie 1992, Steenhof et al. 1997). However, there are little baseline data describing Golden Eagle populations across their range in the western United States (U.S.). The United States Geological Survey (USGS) Snake River Field Station recently prepared a preliminary plan for monitoring Golden Eagle populations (Fuller et al. 2001). Based on recommendations by Fuller et al. (2001), the United States Fish and Wildlife Service (USFWS) issued a request for proposals (Solicitation number 982103R041) to design and conduct Golden Eagle population surveys in Bird Conservation Regions (BCRs) 9,10,16, and 17 within the boundary of the U.S. The overall objective of the project was to estimate Golden Eagle population sizes in the study area using aerial transect procedures that would yield, if replicated annually, at least 80% power to detect an annual rate of total population change greater than or equal to 3 percent per year over a 20-year period using a test of size alpha = 0.1 (or 90% confidence interval).
On July 28, 2003, the USFWS awarded Western EcoSystems Technology (WEST), Inc., (Cheyenne, Wyoming) a contract to design and conduct aerial surveys for Golden Eagles during August and September of 2003, and provide estimates of population sizes within the defined study area. Survey methodology used by WEST, Inc., was based on recommendations by Fuller et al. (2001), with some modifications. We surveyed for Golden Eagles by flying 148 transects, each approximately 100km in length, using three survey crews. Surveys were conducted using Cessna 205 and 206 aircraft flown at approximately 161 km/hr and at either 107m or 150m Above Ground Level (AGL), depending on terrain and safety. At least two observers were present on every survey flight, and a rotating third observer was present on approximately 1/3 of the flights in order to evaluate detection rates (i.e., number of Golden Eagles missed). The surveys were conducted from August 16 – September 8, 2003, after most Golden Eagles had fledged and before fall migration.
A total of 172 Golden Eagles were observed by the survey crews while on transect. We attempted to classify Golden Eagles into one of three age categories, including: adult, older immature (sub-adult), or juvenile. Of the 172 Eagle observations, 58% were classified according to age. Because some perched golden eagles did not provide views of wings and tails, the remaining 42% were classified as unknown adult (adult or older immature), unknown immature (juvenile or older immature) or unknown (no age class assigned).
Using double-observer methods (i.e., use of the third rotating observer), in combination with traditional line transect methodology, we estimated the probability of detecting each Golden Eagle and adjusted our density and abundance estimates accordingly.
We calculated density estimates of Golden Eagles for each of the four BCR’s, for areas surveyed, and a total density for the study area. Within the areas surveyed (areas within 60km of surveyed transects), we estimated a total of 23,012 Golden Eagles (Standard 90% CI: 18,013-29,399). Assuming transect locations were representative of the entire study area, we applied our density estimates to the entire study area, excluding military lands, large water bodies and large urban areas. Under this assumption, we estimated a total of 27,392 Golden Eagles (Standard 90% CI: 21,352-35,140) were present in the study area during the late summer and early fall of 2003. However, this estimate should be considered conservative for two reasons: 1) we did not survey in and extrapolate our estimates to habitat in military owned lands, large urban areas or large bodies of water and 2) we can not adjust the estimates for availability bias on or near the transect line (e.g. those birds that were in the survey strip and on or near the transect line, but hidden from view during surveys).
Due to the difficulty in differentiating between older immature and adult birds in the perched position, we recommend that future surveys include aging all observed Golden Eagles but that trend detection and estimation of yearly status focus on the total population size and the number of juvenile Golden Eagles within the study area.
We investigated two analyses for detecting a change in Golden Eagle numbers, given 20 years of monitoring and various sample sizes (i.e., number of transects). Standard sample size formulas and an extensive computer simulation suggest a minimum sample size of 233 100 km long transects would be necessary for a test of a significant difference in population totals (of 3% per year net change) between years 1 and 20 using a 90% confidence interval, with a power of 80%. Results of our computer simulation also suggested that 150 transects would be sufficient to detect a trend in population size equal to 3 percent per year over a 20-year period using a test on the slope of a regression line, with a test of size alpha = 0.1 (or 90% confidence interval), and at least 80% power. However, the number of double-observer detections acquired by surveying 150 transects would be lower than the recommended sample sizes for generating robust detection functions and visibility correction factors.
Increasing the number of transects flown each survey year will directly result in higher accuracy (lower bias) and precision (smaller variance and CIs) of density and abundance estimates. Additionally, statistical power to detect trends would increase and Type I error rates would be closer to nominal values. We recommend surveying at least 175 100km long transects during future surveys with a double-observer present on every flight. Following the first 3 or 4 years of data collection, proposed sample sizes and sampling intensity could be re-evaluated using data collected during previous surveys. In future analyses, it may also be possible to pool several years of survey data for estimation of detection functions. Our recommended sample sizes and methodologies are based on the 2003 survey data and the stated precision requirements of the USFWS.